Evaluation of China's regional hospital efficiency: DEA approach with undesirable output

Abstract

This paper investigates regional hospital efficiency in China. As hospital operations naturally increase along with an undesirable output of patient mortality, which may induce medical disputes and ligations, this study adopts data envelopment analysis to evaluate and compare the efficiency scores obtained with and without considering the undesirable output. On the basis of province-level panel data over 2002–2008 and considering the risk-adjusted undesirable output, empirical estimates indicate that hospital efficiency is moderate, but increases gradually from 0.6881 to 0.8159. Importantly, without considering the undesirable output, the average efficiency score is overestimated and the efficiency ranking across provinces changes considerably. An efficiency gap is found between coastal and non-coastal regions, but this gap’s drop is mainly contributed by the fast efficiency improvement of the western regions. However, the central regions continue to achieve a significantly lower efficiency score than the eastern and western regions. Moreover, the initiation of the New Rural Cooperative Medical System has overall enhanced hospital efficiency in China, especially for the non-coastal regions.

Keywords:

• hospital
• efficiency
• undesirable output
• DEA

Acknowledgements

The paper is drawn from parts of Hsin-Hui Hu’s Dalian Medical University PhD dissertation. The authors would like to thank two anonymous referees for their insightful comments and suggestions on an earlier version of this paper. The usual disclaimer applies.

Notes

1 For the development of China's urban health insurance system, please see Liu (2002). Alternatively, Liu (2004) and Wang (2007) provide a comprehensive discussion on the evolution of the rural health insurance system. Akin et al (2005), Blomqvist (2009), and Dong (2009) also provide a detailed review on China’s health system reform.

2 In 2007, basic social medical insurance coverage for the urban population and the New Rural Cooperative Medical System (NRCMS) coverage for the rural population is 65.00 and 73.79%, respectively (Dong, 2009).

3 Some studies evaluate health production efficiency by using micro-level data of a small number of hospitals; please refer to Eggleston et al (2008) for a comprehensive survey.

4 O’Neill et al (2008) provide a systematic review of 79 such studies published from 1984 to 2004 that represent 12 countries.

5 Feng and Antony (2010) propose a new methodology that integrates the DEA method into the framework of Six Sigma to assess health service efficiency.

6 Thanassoulis et al (1987) examine the nature of information obtained from DEA in comparative studies of the efficiency of decision-making units, and discuss the interpretation and practical usefulness of such information.

7 Recent studies, such as Färe and Grosskopf (2004) and Seiford and Zhu (2005), have continued to argue with the matter of improving this approach, and it has been widely applied to evaluate efficiency with undesirable outputs, such as banks (Portela et al, 2004) and pollution (Forsund, 2009).

8 There are at least five possibilities for dealing with undesirable outputs in the DEA framework (Seiford and Zhu, 2002).

9 The difference with Seiford and Zhu’s (2002) BCC-DEA framework is that, in the final restriction equation, they pose an addition restriction, to ensure the convexity condition of a VRS model.

10 While part of patient mortality in the hospital results from mistakes on the part of clinicians, mortality can still result even when there are no mistakes, that is, the technical quality of care is excellent. However, due to the lack of information, we cannot construct the variable of risk-adjustment mortality.

11 There are many problems in the Chinese health-care system (Yip and Hsiao, 2008). However, we have no strong attempt and sufficient ability to deal with too many problems in this study. Both aforementioned points are limitations of this study.

12 The measures of inputs and outputs are not the same in both studies, and thus we cannot compare the estimated efficiency scores directly.

13 A more reliable argument regarding the impact of medical reform on hospital efficiency should be based on a longer time span for the pre-reform period.

14 In China, hospitals can apply for ‘class’ certification. The classification system classifies hospitals into third-class (the highest class), second-class, and first-class according to criterions on number of beds, medical departments, personnel, and so on. For details, please see http://www.moh.gov.cn.

15 Zheng and Zhang (2009) propose some possible interpretations.

16 The Mid-China regions include: Hebei, Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, and Hainan. The West China regions contain Neimenggu, Guangxi, Chongqing, Sichun, Guizhou, Yunan, Shangxi, Gansui, Qinghai, and Tibet.

17 The main components of the strategy include development of infrastructure, enticement of foreign investment, increased efforts on ecological protection, promotion of education, and retention of talent flowing to richer provinces. For the details of this programme, please see http://www.westchina.gov.cn.

18 In fact, the eastern region is a subsample of the coastal region that excludes Hebei, Guangxi, and Hainan.

19 Indeed, there are many potential confounders that could cause one region to be more efficient than another, including labor costs, local diet, population health factors, environmental factors, other economic factors. These factors may induce the manager of a hospital or a group of hospitals to choose the hospital's location and further impact on hospital efficiency. This issue is worth further investigation.